Monetary Economics: Macro Aspects, 2/4 2013 Henrik Jensen Department of Economics University of Copenhagen Monetary credibility problems 1. In ation and discretionary monetary policy 2. Reputational solution to credibility problems Literature: Walsh (Chapter 7, pp. 269-290) Note a small accompanying note on web, as well as an addendum at the end of slides on further solutions to credibility problems c 2013 Henrik Jensen. This document may be reproduced for educational and research purposes, as long as the copies contain this notice and are retained for personal use or distributed free.
Introductory remarks So far, we have treated monetary policy as some exogenous process Is this an innocent assumption? No, as one has not speci ed how the policymaker can be guaranteed to adhere to such rules I.e., what are the policymaker s incentives? What if the policymaker responds to incentives? What if the policymaker under some circumstances has the incentive to conduct policy that di ers from the rule? This may feed back into private sector behavior, as this is based on expectations and thus expected policy It therefore becomes of importance whether the policy rule is credible If it is not credible, private sector expectations will adjust towards what is believed the policymaker will do instead of the rule 1
It therefore becomes important to understand the policymaker s incentives, as this will determine whether it will adhere to the rule, or act discretionary This, in turn, provides an understanding about the determination of private sector expectations and thus the macroeconomic equilibrium General insight in frameworks where private sector expectations are important: Policies that are optimal ex ante, are often not optimal ex post They are time inconsistent (Kydland and Prescott, 1977) => good rules may not be credible 2
This implies that the policymaker often have an incentive to deviate from the optimal rule when private sector expectations are formed This is then taken into account in expectations formation, and the economy may end up in an equilibrium which is time consistent (a discretionary equilibrium), but suboptimal: Given the expectations that incorporate the policymaker s incentive to deviate from the optimal rule...... monetary policy end up delivering a poor result This is the positive/descriptive aspect of time inconsistency models If this is important for various policy scenarios, it raises the issue: How can the policymaker be endowed with commitment abilities, which overcome the timeinconsistency problem of the optimal plan? How can society though legislation, or institutional design, change the policymaker s incentives such that the poor macroeconomic equilibrium is avoided? This is the normative aspect of the time inconsistency models 3
In ation and discretionary monetary policy The Barro Gordon model exempli es how optimal monetary policy can be time inconsistent, and demonstrates the negative consequences on the economy The Barro-Gordon in ation bias model Economic structure is a very simple AS/AD setting AS curve: y t = y n + a ( t e t) + e t ; a > 0 (7.3) Simple price/in ation surprise equation (in logs) AD curve: t = m t + v t (7.4) Further simpli cation: t = m t (7.4 ) In ation rate is taken as the monetary policy instrument 4
Two variants of the model di ering in terms of central bank s preferences: Variant 1. Utility linear in output: Ut 1 1 = (y t y n ) 2 2 t; > 0 (7.1) The central bank wishes output to exceed the natural rate as much as possible, but dislikes deviations in in ation from zero (normalization of in ation goal) Variant 2. Utility quadratic in output The central bank: U 2 t = 2 (y t y n k) 2 1 2 2 t; k > 0; (7.2) dislikes deviations in output from y n + k > y n dislikes deviations in in ation from zero. Note that preferred output level is higher than the natural rate as k > 0 5
Under both variants, the timing of events within the period is: 1: e t is formed 2: e t is realized 3: t is set 4: y t is determined Central features of this move structure: Private sector is committed to a xed in ation expectation as nominal wage contracts are written in the beginning of the period The private sector does not know e t at the time they sign contracts; hence, e t is not conditioned on e t The central bank performs policy after having observed the shock, and after in ation expectations are formed This makes unanticipated in ation leading to output e ects possible even in absence of shocks However, it is assumed that the private sector knows the incentives of the central bank, which is crucial for the determination of e t It is assumed that e t = E t 1 t 6
Solution of model under Variant 1 Solution concept is time-consistent Nash equilibrium Nash equilibrium: Agents choices are best responses to other agents choices Time consistency: Agents choices as prescribed by the solution will not be changed later To attain time consistency, one solves model by backward induction First, determine the optimal monetary policy, for given expectations and supply shock: max U 1 1 t = (y t y n ) t 2 2 t First-order condition: = (y n + a ( t e t) + e t y n ) Marginal gain of in ation equals the marginal loss 1 2 2 t = [a ( t e t) + e t ] 1 2 2 t a = t (7.5 ) This readily gives the solution for monetary policy. In ation expectations are found to be the same, as the private sector can foresee the policy as given by (7.5 ). I.e., e t = E t 1 t = a Solution for output: y t = y n + e t 7
Equilibrium features a positive rate of in ation an in ation bias which has no e ect on output, since it is perfectly anticipated by the public Why? Because the private sector understands the central bank s incentive to raise output by creating an in ation surprise. The private sector incorporates this into expectations, and in ation becomes ine ciently high The time-inconsistency of the optimal solution The central bank would be better o setting t = 0. If this is credible, e t become y t = y n + e t. = 0, and output will Problem: This is time-inconsistent. If e t = 0, it is optimal to set t = a driving output above the natural rate, y t = a 2 + y n + e t The private sector foresees this incentive, and will not set e t = 0 in the rst place...only e t = t = a is a time-consistent solution A simple, but strong, example of how a well-meaning policymaker without credibility, i.e., without an ability to commit, brings the economy into a bad equilibrium The discretionary nature of monetary policy brings about excessive in ation 8
Solution of model under Variant 2 Main message applies, but policy will now respond to the supply shock When policy is implemented, in ation expectations and the supply shock are taken as given, and the central bank solves max Ut 2 = t 2 (y t y n k) 2 1 2 2 t = 2 [a ( t e t) + e t k] 2 1 2 2 t The rst-order condition is a [a ( t e t) + e t k] = t Marginal bene t of in ation in terms of output equals the marginal loss in terms of in ation per se Reaction function becomes t = a 1 + a 2 (ae t e t + k) (7.6 ) Higher in ation expectations drive up in ation, as the marginal gain in terms of higher output increases A positive supply shock is met by a contractive monetary policy, as stabilization of output now is a concern The more the output goal exceeds the natural rate (the higher k) the higher in ation 9
The private sector can foresee this reaction function, but doesn t know the value of the supply shock. Assuming a mean-zero shock, in ation expectations are found from implying Actual discretionary in ation becomes: Actual discretionary output becomes: e t = a 1 + a 2 (ae t + k) e t = ak > 0 t = ak y t = y n + a 1 + a 2e t (7.7 ) 1 1 + a 2e t Again, an in ation bias prevails with no e ect on output. If commitment to a policy rule t = b 0 b 1 e t was possible, the optimal values of coe cients are (see note on web) b 0 = 0; b 1 = a 1 + a 2 Stabilization property of discretionary policy is optimal, but the average is ine cient due to the in ation bias 10
Earlier debates (pre 1980s) about rules versus discretion in monetary policy would end up favoring discretion, if there was a need and scope for stabilization against economic shocks With time-inconsistency problems this may not be the case. A rigid rule that does not respond to shocks, t = 0, may be desirable in the Barro and Gordon model under Variant 2...... if the variability of e t is not too large See condition on p. 282 in Walsh 11
Reputational solutions to credibility problems Is it impossible to avoid the excessive in ation result? No. For the remainder we consider a market-based solution (as opposed to more institutional based solutions mentioned at the end): Namely one of reputation building Interactions between the private sector and policymakers are rarely of a one-shot nature. They occur repeatedly, and this opens the possibility for reputation building This can be modelled in a simple manner. Assume that the interaction between the private sector and the central bank, the game, is repeated in nitely Then, doing bad today, may cause loss of reputation tomorrow that in e ect prevents the central bank from doing bad today For this to be the case, the private sector must punish bad behavior of the central bank; i.e., punish a deviation from a promised in ation rate Assume there are no supply shocks, and Variation 1 utility applies 12
The central bank promises < a. The potential punishment behavior by the private sector is modelled by a simple trigger strategy : e t = < a if t 1 = e t 1 e t = a otherwise I.e., if the central bank did not surprise in the previous period the public keeps expecting the promise,. If not, it reverts to the bad equilibrium in ation expectations for one period Tit for tat strategy in game-theoretic language Note that this is one of in nitely many punishment strategies (how can private sector agents coordinate on a particular one? issue is ignored here, where purpose is to show that it is possible to obtain a reputation for low in ation) The central bank now makes decisions knowing that breaking a promise has implications for future private sector behavior The central bank s objective is to maximize the discounted sum of per-period utilities: 1X max t Ut 1 ; 0 < < 1 Which in ation rates, if any, can the central bank get a reputation for delivering? t=0 13
What is the optimal deviation from in a period? From before: t = a. Associated current net gain temptation for the central bank: 1 1 a (a ) {z 2 (a)2 + } 2 ()2 = 1 (a )2 {z } 2 Utility from surprise Utility from promise G () What is optimal policy under the punishment e t = a? From before: t = a. Associated future net loss enforcement for the central bank following a deviation: 2 {z ()2 + } 2 (a)2 = h(a) 2 () 2i {z } 2 Utility from promise Utility from punishment C () (7.13) An in ation rate policy promise,, is credible if C () G (); i.e., if the temptation is not stronger than the enforcement. This implies 2 h (a) 2 () 2i 1 (a )2 2 This means that any policy is credible in the repeated game if a 1 1 + a Note that the minimum sustainable in ation rate is decreasing in (and for! 1, it goes to zero). 14
Summary and addendum Time-inconsistency problems of optimal monetary policy create suboptimal outcomes when the central bank cannot credibly commit The Barro and Gordon model provides a simple example of the basic nature of such credibility problems Credibility problems in monetary policymaking (and other branches of policymaking) have costs which can be identi ed by economic models More importantly, models of credibility problems provide a natural platform for start thinking about how society can overcome credibility problems and their associated costs Either through some form of reputation building...... or through establishment of economic institutions which shape the policymakers incentives in a direction that mitigates the incentives to deviate from the optimal plan; often referred to ad delegation theories Of these, the Rogo central banker, Walsh contracts and In ation targeting have received lots of attention. See Walsh (Chapter 7, pp. 297 323) and ensuing addendum 15
Addendum: Delegation and independent central banks The incentive to surprise in ation is often interpreted as arising from political pressures Solution to time-inconsistency problem could be achieved by Delegating monetary policy conduct to independent central banks Create monetary institutions securing independence and appropriate policy incentives I.e., appropriate design of policy regime in broadest sense Analyses are cast in versions of Barro Gordon model with Variant 2 utility Under delegation, the institutional design stage is added to move structure: 1: Establishment of monetary delegation regime 2: e is formed 3: e is realized 4: is set 5: y is determined 16
Delegation to a conservative central banker The idea is to appoint a central banker, who puts relative more weight on in ation stabilization than society I.e., monetary policy is delegated to central banker with utility U c = 2 (y y n k) 2 1 + 2 2 ; > 0 This is Rogo s conservative central banker; measures the degree of conservativeness (Rogo, 1985, QJE) Monetary policymaking by the central banker (taking as given e and e) is characterized by the rst-order condition a [a ( e ) + e k] = (1 + ) (*) Note that > 0 increases the marginal cost of in ation Rational in ation expectations follow by taking expectations on both sides of (*): ak = (1 + ) E [] =) E [] = ak 1 + < ak 17
With a conservative central banker, the in ation bias is reduced from ak to ak= (1 + ) Conservativeness, however, has a cost. The solution for actual in ation becomes [plug the solution for e = E[] back into (*)] = ak a 1 + 1 + + a 2e Stabilization of the shock is distorted Compared to the socially optimal response to a supply shock, a conservative central bank responds less to the shock Result is too stable in ation and too unstable output Appointing a conservative central bank thus involves a trade-o between a) Lower average in ation b) Poorer macroeconomic stabilization So, will it ever be optimal to have > 0? Yes, always At = 0, a marginal increase in involves a rst-order social gain of lower average in ation (at = 0 average in ation is suboptimal) At = 0, a marginal increase in involves a second-order social loss of poorer stabilization (at = 0 stabilization is optimal) 18
Incentive contracts Under this approach, the government appoints a central bank, and o ers him/her a performance contract This contract rewards or punishes the central bank depending on its performance The contract could be pecuniary but more generally, it could represent public embarrassment if the central bank doesn t ful ll its contract Real world analogy: The Federal Reserve Act of 1989 in New Zealand: The governor can be red, if he performs poorly; other central bankers are in other ways held accountable for their actions. Formally, the central bank is o ered a contract, such that it maximizes where t is the contract transfer U + t 19
Assume that the contract transfer cannot be made contingent on the supply shock, and only a transfer depending on observed in ation is considered: t = t () Task of government is to choose the optimal t () (at institutional design stage) Central bank takes expectations and the supply shock as given, and maximizes The rst-order condition is 2 (a ( e ) + e k) 2 1 2 2 + t () a [a ( e ) + e k] = t 0 () (**) If t 0 () < 0 we see that the marginal cost of in ation is higher than without the transfer; i.e., the contract punishes in ation increases Rational in ation expectations follow by taking expectations on both sides of (**): ak = E [] E [t 0 ()] =) E [] = ak + E [t 0 ()] 20
Insert these expectations back into (**) to get actual in ation a (a ( ak E [t 0 ()]) + e k) = t 0 () = ak + a2 1 + a 2E [t0 ()] + t0 () a 1 + a 2 1 + a 2e Optimal policy is implemented if the transfer function satis es ak + a2 1 + a 2E [t0 ()] + t0 () 1 + a = 0 2 This is accomplished if A transfer function with this property: A linear in ation contract t 0 () = t () = t 0 ak ak Linear because the incentive to surprise the private sector is a constant in equilibrium; hence, a constant marginal punishment eliminates the in ation bias (also for non-quadratic utility) In contrast to a conservative central banker, the linear in ation contract portrays the optimal incentive structure 21
Plan for next lectures Tuesday, April 9, Exercises: Exercise posted on web. Thursday, April 11, Lectures: 1. Operating procedures and choice of monetary policy instrument 2. Intermediate targets in policymaking Literature: Walsh (Chapter 11, pp. 512 530) 22